numberyFuzzyMatcher.py

from fuzzywuzzy import fuzz
import re

def partialNumberyRatio(stringA, stringB):
    stringA = stringA.lower()
    stringB = stringB.lower()

    digitsA = "".join(sorted([char for char in stringA if char.isdigit()]))
    digitsB = "".join(sorted([char for char in stringB if char.isdigit()]))
    
    
    digitMatch = float(digitsA in digitsB)


    match = fuzz.partial_ratio(stringA, stringB)    
    
    return match * digitMatch



print partialNumberyRatio("sweata", "gatorade_SweatMachine_4592")
print partialNumberyRatio("010", "sh010")
print partialNumberyRatio("010", "sh080")
print partialNumberyRatio("010", "sh020")
print partialNumberyRatio("010", "sh100")
print partialNumberyRatio("gfx010 1 600", "GVFX010_P01_060")
print partialNumberyRatio("10 1 600", "GVFX010_P01_060")

numberyFuzzyMatcher2.py

from fuzzywuzzy import fuzz
import re


def findTokens(string):
    """split string into two sets of tokens, numbers and letters. preserves length by filling gaps with None"""
    groups = [match.groupdict() for match in re.finditer("(?P<d>\d+)|(?P<c>[A-Za-z]+)", string)]
    # return [group.get("d") or group.get("c") for group in groups]
    return tuple([group.get("d") for group in groups]), tuple([group.get("c") for group in groups])


def findNumbers(string):
    """get list of all numbers in a string"""
    return re.findall("\d+", string)


def partialNumberyRatio(stringA, stringB):
    stringA = stringA.lower()
    stringB = stringB.lower()

    aNums = "".join(findNumbers(stringA))
    numbersMatch = inStringInOrder(aNums, stringB)

    textMatch = fuzz.partial_ratio(stringA, stringB)
    print stringA, stringB, int(textMatch * numbersMatch)
    return int(textMatch * numbersMatch)

def inStringInOrder(search, string):
    """@return: True if the characters in search exist in order in string"""
    query = ".*".join(search)
    return not (re.search(query, string) is None)